The memory slot management functions were oriented against vcpu 0, where
they should be kvm-wide. This causes hangs starting X on guest smp.
Fix by making the functions (and resultant tail in the mmu) non-vcpu-specific.
Unfortunately this reduces the efficiency of the mmu object cache a bit. We
may have to revisit this later.
Signed-off-by: Avi Kivity <[email protected]>
---
drivers/kvm/kvm.h | 4 +-
drivers/kvm/kvm_main.c | 68 ++++---------------------------
drivers/kvm/mmu.c | 103 +++++++++++++++++++-----------------------------
3 files changed, 52 insertions(+), 123 deletions(-)
diff --git a/drivers/kvm/kvm.h b/drivers/kvm/kvm.h
index 65ab268..6636ae2 100644
--- a/drivers/kvm/kvm.h
+++ b/drivers/kvm/kvm.h
@@ -535,8 +535,8 @@ int kvm_mmu_create(struct kvm_vcpu *vcpu);
int kvm_mmu_setup(struct kvm_vcpu *vcpu);
int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
-void kvm_mmu_slot_remove_write_access(struct kvm_vcpu *vcpu, int slot);
-void kvm_mmu_zap_all(struct kvm_vcpu *vcpu);
+void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
+void kvm_mmu_zap_all(struct kvm *kvm);
hpa_t gpa_to_hpa(struct kvm_vcpu *vcpu, gpa_t gpa);
#define HPA_MSB ((sizeof(hpa_t) * 8) - 1)
diff --git a/drivers/kvm/kvm_main.c b/drivers/kvm/kvm_main.c
index 1b206f1..05f0418 100644
--- a/drivers/kvm/kvm_main.c
+++ b/drivers/kvm/kvm_main.c
@@ -238,23 +238,6 @@ static void vcpu_load(struct kvm_vcpu *vcpu)
kvm_arch_ops->vcpu_load(vcpu);
}
-/*
- * Switches to specified vcpu, until a matching vcpu_put(). Will return NULL
- * if the slot is not populated.
- */
-static struct kvm_vcpu *vcpu_load_slot(struct kvm *kvm, int slot)
-{
- struct kvm_vcpu *vcpu = &kvm->vcpus[slot];
-
- mutex_lock(&vcpu->mutex);
- if (!vcpu->vmcs) {
- mutex_unlock(&vcpu->mutex);
- return NULL;
- }
- kvm_arch_ops->vcpu_load(vcpu);
- return vcpu;
-}
-
static void vcpu_put(struct kvm_vcpu *vcpu)
{
kvm_arch_ops->vcpu_put(vcpu);
@@ -663,13 +646,6 @@ void fx_init(struct kvm_vcpu *vcpu)
}
EXPORT_SYMBOL_GPL(fx_init);
-static void do_remove_write_access(struct kvm_vcpu *vcpu, int slot)
-{
- spin_lock(&vcpu->kvm->lock);
- kvm_mmu_slot_remove_write_access(vcpu, slot);
- spin_unlock(&vcpu->kvm->lock);
-}
-
/*
* Allocate some memory and give it an address in the guest physical address
* space.
@@ -792,19 +768,10 @@ raced:
*memslot = new;
++kvm->memory_config_version;
- spin_unlock(&kvm->lock);
-
- for (i = 0; i < KVM_MAX_VCPUS; ++i) {
- struct kvm_vcpu *vcpu;
+ kvm_mmu_slot_remove_write_access(kvm, mem->slot);
+ kvm_flush_remote_tlbs(kvm);
- vcpu = vcpu_load_slot(kvm, i);
- if (!vcpu)
- continue;
- if (new.flags & KVM_MEM_LOG_DIRTY_PAGES)
- do_remove_write_access(vcpu, mem->slot);
- kvm_mmu_reset_context(vcpu);
- vcpu_put(vcpu);
- }
+ spin_unlock(&kvm->lock);
kvm_free_physmem_slot(&old, &new);
return 0;
@@ -826,7 +793,6 @@ static int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm,
struct kvm_memory_slot *memslot;
int r, i;
int n;
- int cleared;
unsigned long any = 0;
spin_lock(&kvm->lock);
@@ -855,23 +821,11 @@ static int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm,
if (copy_to_user(log->dirty_bitmap, memslot->dirty_bitmap, n))
goto out;
- if (any) {
- cleared = 0;
- for (i = 0; i < KVM_MAX_VCPUS; ++i) {
- struct kvm_vcpu *vcpu;
-
- vcpu = vcpu_load_slot(kvm, i);
- if (!vcpu)
- continue;
- if (!cleared) {
- do_remove_write_access(vcpu, log->slot);
- memset(memslot->dirty_bitmap, 0, n);
- cleared = 1;
- }
- kvm_arch_ops->tlb_flush(vcpu);
- vcpu_put(vcpu);
- }
- }
+ spin_lock(&kvm->lock);
+ kvm_mmu_slot_remove_write_access(kvm, log->slot);
+ kvm_flush_remote_tlbs(kvm);
+ memset(memslot->dirty_bitmap, 0, n);
+ spin_unlock(&kvm->lock);
r = 0;
@@ -920,13 +874,9 @@ static int kvm_vm_ioctl_set_memory_alias(struct kvm *kvm,
break;
kvm->naliases = n;
- spin_unlock(&kvm->lock);
+ kvm_mmu_zap_all(kvm);
- vcpu_load(&kvm->vcpus[0]);
- spin_lock(&kvm->lock);
- kvm_mmu_zap_all(&kvm->vcpus[0]);
spin_unlock(&kvm->lock);
- vcpu_put(&kvm->vcpus[0]);
return 0;
diff --git a/drivers/kvm/mmu.c b/drivers/kvm/mmu.c
index b297a6b..970f1e5 100644
--- a/drivers/kvm/mmu.c
+++ b/drivers/kvm/mmu.c
@@ -281,24 +281,15 @@ static void *mmu_memory_cache_alloc(struct kvm_mmu_memory_cache *mc,
return p;
}
-static void mmu_memory_cache_free(struct kvm_mmu_memory_cache *mc, void *obj)
-{
- if (mc->nobjs < KVM_NR_MEM_OBJS)
- mc->objects[mc->nobjs++] = obj;
- else
- kfree(obj);
-}
-
static struct kvm_pte_chain *mmu_alloc_pte_chain(struct kvm_vcpu *vcpu)
{
return mmu_memory_cache_alloc(&vcpu->mmu_pte_chain_cache,
sizeof(struct kvm_pte_chain));
}
-static void mmu_free_pte_chain(struct kvm_vcpu *vcpu,
- struct kvm_pte_chain *pc)
+static void mmu_free_pte_chain(struct kvm_pte_chain *pc)
{
- mmu_memory_cache_free(&vcpu->mmu_pte_chain_cache, pc);
+ kfree(pc);
}
static struct kvm_rmap_desc *mmu_alloc_rmap_desc(struct kvm_vcpu *vcpu)
@@ -307,10 +298,9 @@ static struct kvm_rmap_desc *mmu_alloc_rmap_desc(struct kvm_vcpu *vcpu)
sizeof(struct kvm_rmap_desc));
}
-static void mmu_free_rmap_desc(struct kvm_vcpu *vcpu,
- struct kvm_rmap_desc *rd)
+static void mmu_free_rmap_desc(struct kvm_rmap_desc *rd)
{
- mmu_memory_cache_free(&vcpu->mmu_rmap_desc_cache, rd);
+ kfree(rd);
}
/*
@@ -355,8 +345,7 @@ static void rmap_add(struct kvm_vcpu *vcpu, u64 *spte)
}
}
-static void rmap_desc_remove_entry(struct kvm_vcpu *vcpu,
- struct page *page,
+static void rmap_desc_remove_entry(struct page *page,
struct kvm_rmap_desc *desc,
int i,
struct kvm_rmap_desc *prev_desc)
@@ -376,10 +365,10 @@ static void rmap_desc_remove_entry(struct kvm_vcpu *vcpu,
prev_desc->more = desc->more;
else
set_page_private(page,(unsigned long)desc->more | 1);
- mmu_free_rmap_desc(vcpu, desc);
+ mmu_free_rmap_desc(desc);
}
-static void rmap_remove(struct kvm_vcpu *vcpu, u64 *spte)
+static void rmap_remove(u64 *spte)
{
struct page *page;
struct kvm_rmap_desc *desc;
@@ -407,7 +396,7 @@ static void rmap_remove(struct kvm_vcpu *vcpu, u64 *spte)
while (desc) {
for (i = 0; i < RMAP_EXT && desc->shadow_ptes[i]; ++i)
if (desc->shadow_ptes[i] == spte) {
- rmap_desc_remove_entry(vcpu, page,
+ rmap_desc_remove_entry(page,
desc, i,
prev_desc);
return;
@@ -442,7 +431,7 @@ static void rmap_write_protect(struct kvm_vcpu *vcpu, u64 gfn)
BUG_ON(!(*spte & PT_PRESENT_MASK));
BUG_ON(!(*spte & PT_WRITABLE_MASK));
rmap_printk("rmap_write_protect: spte %p %llx\n", spte, *spte);
- rmap_remove(vcpu, spte);
+ rmap_remove(spte);
set_shadow_pte(spte, *spte & ~PT_WRITABLE_MASK);
kvm_flush_remote_tlbs(vcpu->kvm);
}
@@ -464,14 +453,14 @@ static int is_empty_shadow_page(u64 *spt)
}
#endif
-static void kvm_mmu_free_page(struct kvm_vcpu *vcpu,
+static void kvm_mmu_free_page(struct kvm *kvm,
struct kvm_mmu_page *page_head)
{
ASSERT(is_empty_shadow_page(page_head->spt));
list_del(&page_head->link);
- mmu_memory_cache_free(&vcpu->mmu_page_cache, page_head->spt);
- mmu_memory_cache_free(&vcpu->mmu_page_header_cache, page_head);
- ++vcpu->kvm->n_free_mmu_pages;
+ kfree(page_head->spt);
+ kfree(page_head);
+ ++kvm->n_free_mmu_pages;
}
static unsigned kvm_page_table_hashfn(gfn_t gfn)
@@ -537,8 +526,7 @@ static void mmu_page_add_parent_pte(struct kvm_vcpu *vcpu,
pte_chain->parent_ptes[0] = parent_pte;
}
-static void mmu_page_remove_parent_pte(struct kvm_vcpu *vcpu,
- struct kvm_mmu_page *page,
+static void mmu_page_remove_parent_pte(struct kvm_mmu_page *page,
u64 *parent_pte)
{
struct kvm_pte_chain *pte_chain;
@@ -565,7 +553,7 @@ static void mmu_page_remove_parent_pte(struct kvm_vcpu *vcpu,
pte_chain->parent_ptes[i] = NULL;
if (i == 0) {
hlist_del(&pte_chain->link);
- mmu_free_pte_chain(vcpu, pte_chain);
+ mmu_free_pte_chain(pte_chain);
if (hlist_empty(&page->parent_ptes)) {
page->multimapped = 0;
page->parent_pte = NULL;
@@ -643,7 +631,7 @@ static struct kvm_mmu_page *kvm_mmu_get_page(struct kvm_vcpu *vcpu,
return page;
}
-static void kvm_mmu_page_unlink_children(struct kvm_vcpu *vcpu,
+static void kvm_mmu_page_unlink_children(struct kvm *kvm,
struct kvm_mmu_page *page)
{
unsigned i;
@@ -655,10 +643,10 @@ static void kvm_mmu_page_unlink_children(struct kvm_vcpu *vcpu,
if (page->role.level == PT_PAGE_TABLE_LEVEL) {
for (i = 0; i < PT64_ENT_PER_PAGE; ++i) {
if (pt[i] & PT_PRESENT_MASK)
- rmap_remove(vcpu, &pt[i]);
+ rmap_remove(&pt[i]);
pt[i] = 0;
}
- kvm_flush_remote_tlbs(vcpu->kvm);
+ kvm_flush_remote_tlbs(kvm);
return;
}
@@ -669,19 +657,18 @@ static void kvm_mmu_page_unlink_children(struct kvm_vcpu *vcpu,
if (!(ent & PT_PRESENT_MASK))
continue;
ent &= PT64_BASE_ADDR_MASK;
- mmu_page_remove_parent_pte(vcpu, page_header(ent), &pt[i]);
+ mmu_page_remove_parent_pte(page_header(ent), &pt[i]);
}
- kvm_flush_remote_tlbs(vcpu->kvm);
+ kvm_flush_remote_tlbs(kvm);
}
-static void kvm_mmu_put_page(struct kvm_vcpu *vcpu,
- struct kvm_mmu_page *page,
+static void kvm_mmu_put_page(struct kvm_mmu_page *page,
u64 *parent_pte)
{
- mmu_page_remove_parent_pte(vcpu, page, parent_pte);
+ mmu_page_remove_parent_pte(page, parent_pte);
}
-static void kvm_mmu_zap_page(struct kvm_vcpu *vcpu,
+static void kvm_mmu_zap_page(struct kvm *kvm,
struct kvm_mmu_page *page)
{
u64 *parent_pte;
@@ -697,15 +684,15 @@ static void kvm_mmu_zap_page(struct kvm_vcpu *vcpu,
parent_pte = chain->parent_ptes[0];
}
BUG_ON(!parent_pte);
- kvm_mmu_put_page(vcpu, page, parent_pte);
+ kvm_mmu_put_page(page, parent_pte);
set_shadow_pte(parent_pte, 0);
}
- kvm_mmu_page_unlink_children(vcpu, page);
+ kvm_mmu_page_unlink_children(kvm, page);
if (!page->root_count) {
hlist_del(&page->hash_link);
- kvm_mmu_free_page(vcpu, page);
+ kvm_mmu_free_page(kvm, page);
} else
- list_move(&page->link, &vcpu->kvm->active_mmu_pages);
+ list_move(&page->link, &kvm->active_mmu_pages);
}
static int kvm_mmu_unprotect_page(struct kvm_vcpu *vcpu, gfn_t gfn)
@@ -724,7 +711,7 @@ static int kvm_mmu_unprotect_page(struct kvm_vcpu *vcpu, gfn_t gfn)
if (page->gfn == gfn && !page->role.metaphysical) {
pgprintk("%s: gfn %lx role %x\n", __FUNCTION__, gfn,
page->role.word);
- kvm_mmu_zap_page(vcpu, page);
+ kvm_mmu_zap_page(vcpu->kvm, page);
r = 1;
}
return r;
@@ -737,7 +724,7 @@ static void mmu_unshadow(struct kvm_vcpu *vcpu, gfn_t gfn)
while ((page = kvm_mmu_lookup_page(vcpu, gfn)) != NULL) {
pgprintk("%s: zap %lx %x\n",
__FUNCTION__, gfn, page->role.word);
- kvm_mmu_zap_page(vcpu, page);
+ kvm_mmu_zap_page(vcpu->kvm, page);
}
}
@@ -1089,10 +1076,10 @@ static void mmu_pte_write_zap_pte(struct kvm_vcpu *vcpu,
pte = *spte;
if (is_present_pte(pte)) {
if (page->role.level == PT_PAGE_TABLE_LEVEL)
- rmap_remove(vcpu, spte);
+ rmap_remove(spte);
else {
child = page_header(pte & PT64_BASE_ADDR_MASK);
- mmu_page_remove_parent_pte(vcpu, child, spte);
+ mmu_page_remove_parent_pte(child, spte);
}
}
*spte = 0;
@@ -1161,7 +1148,7 @@ void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
*/
pgprintk("misaligned: gpa %llx bytes %d role %x\n",
gpa, bytes, page->role.word);
- kvm_mmu_zap_page(vcpu, page);
+ kvm_mmu_zap_page(vcpu->kvm, page);
continue;
}
page_offset = offset;
@@ -1207,7 +1194,7 @@ void kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu)
page = container_of(vcpu->kvm->active_mmu_pages.prev,
struct kvm_mmu_page, link);
- kvm_mmu_zap_page(vcpu, page);
+ kvm_mmu_zap_page(vcpu->kvm, page);
}
}
EXPORT_SYMBOL_GPL(kvm_mmu_free_some_pages);
@@ -1219,7 +1206,7 @@ static void free_mmu_pages(struct kvm_vcpu *vcpu)
while (!list_empty(&vcpu->kvm->active_mmu_pages)) {
page = container_of(vcpu->kvm->active_mmu_pages.next,
struct kvm_mmu_page, link);
- kvm_mmu_zap_page(vcpu, page);
+ kvm_mmu_zap_page(vcpu->kvm, page);
}
free_page((unsigned long)vcpu->mmu.pae_root);
}
@@ -1277,9 +1264,8 @@ void kvm_mmu_destroy(struct kvm_vcpu *vcpu)
mmu_free_memory_caches(vcpu);
}
-void kvm_mmu_slot_remove_write_access(struct kvm_vcpu *vcpu, int slot)
+void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot)
{
- struct kvm *kvm = vcpu->kvm;
struct kvm_mmu_page *page;
list_for_each_entry(page, &kvm->active_mmu_pages, link) {
@@ -1293,27 +1279,20 @@ void kvm_mmu_slot_remove_write_access(struct kvm_vcpu *vcpu, int slot)
for (i = 0; i < PT64_ENT_PER_PAGE; ++i)
/* avoid RMW */
if (pt[i] & PT_WRITABLE_MASK) {
- rmap_remove(vcpu, &pt[i]);
+ rmap_remove(&pt[i]);
pt[i] &= ~PT_WRITABLE_MASK;
}
}
}
-void kvm_mmu_zap_all(struct kvm_vcpu *vcpu)
+void kvm_mmu_zap_all(struct kvm *kvm)
{
- destroy_kvm_mmu(vcpu);
+ struct kvm_mmu_page *page, *node;
- while (!list_empty(&vcpu->kvm->active_mmu_pages)) {
- struct kvm_mmu_page *page;
-
- page = container_of(vcpu->kvm->active_mmu_pages.next,
- struct kvm_mmu_page, link);
- kvm_mmu_zap_page(vcpu, page);
- }
+ list_for_each_entry_safe(page, node, &kvm->active_mmu_pages, link)
+ kvm_mmu_zap_page(kvm, page);
- mmu_free_memory_caches(vcpu);
- kvm_flush_remote_tlbs(vcpu->kvm);
- init_kvm_mmu(vcpu);
+ kvm_flush_remote_tlbs(kvm);
}
void kvm_mmu_module_exit(void)
--
1.5.2.4
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Stuff]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
[Linux Resources]
